AMD Radeon Instinct MI60 vs AMD Radeon RX 9070 GRE Comparison

Where Each One Wins

The benchmark data splits these two AMD cards into very different roles. The Radeon Instinct MI60, a data center compute card from 2018, is built around massive memory bandwidth and compute throughput. The Radeon RX 9070 GRE, a 2025 gaming-oriented part, focuses on raw shader output and modern API support. Looking at the recorded measurements, the RX 9070 GRE wins the only direct head-to-head test available, scoring 109309 in Geekbench OpenCL against the MI60's 92488, a 15.4% advantage. That single win, however, masks the broader picture: the MI60 sits at the 93rd percentile among all GPUs in the database, while the RX 9070 GRE is at the 87th percentile. The MI60's average benchmark score of 92466 is actually higher than the RX 9070 GRE's average of 57367, though the latter figure is dragged down by including a 3DMark Steel Nomad DX12 result of 5424, a test the MI60 was never run on.

The MI60 wins in scenarios that favor its HBM2 memory and 4096-bit bus. Its 1.02 TB/s bandwidth is more than double the RX 9070 GRE's 432.0 GB/s, making it the obvious choice for workloads that stream large datasets, such as scientific simulation or machine learning inference where data movement dominates. Its FP16 throughput of 29.49 TFLOPS (2:1) also gives it a compute edge in mixed-precision tasks, even though its FP32 of 14.75 TFLOPS is less than half the RX 9070 GRE's 34.28 TFLOPS. The MI60's 32 GB of memory is a decisive advantage for models or buffers that exceed 12 GB, and its 4096-bit bus means memory-bound kernels see far less contention.

The RX 9070 GRE wins in almost every other measurable category. Its 34.28 TFLOPS FP32 is 132% higher than the MI60's, and its FP16 is also 34.28 TFLOPS (1:1), meaning it does not trade off precision for speed. Its pixel rate of 267.8 GPixel/s is 132% higher than the MI60's 115.2 GPixel/s, and its texture rate of 535.7 GTexel/s beats the MI60's 460.8 GTexel/s by 16%. It also has 48 dedicated ray tracing cores, which the MI60 lacks entirely. For gaming, the RX 9070 GRE supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, while the MI60 is limited to DirectX 12 (12_1) and Vulkan 1.3. The RX 9070 GRE's boost clock of 2790 MHz is 55% higher than the MI60's 1800 MHz, and its base clock of 1420 MHz is also higher. This is a card designed to push frames, not just numbers.

Architecture Differences

The two GPUs come from entirely different architectural eras. The MI60 uses the Vega 20 chip based on GCN 5.1, a compute-first design from 2018. It is built on TSMC's 7 nm process with 13,230 million transistors on a 331 mm² die, giving a transistor density of 40.0 million per mm². The RX 9070 GRE uses the Navi 48 chip based on RDNA 4.0, a gaming-focused architecture from 2025. It is built on TSMC's 4 nm process with 53,900 million transistors on a 357 mm² die, yielding 151.0 million transistors per mm², a 3.8x density improvement. That density advantage allows the RX 9070 GRE to pack more compute into a similar physical area.

The memory subsystems are fundamentally different. The MI60 uses 32 GB of HBM2 with a 4096-bit bus and 1.02 TB/s bandwidth, a design aimed at data center workloads where capacity and bandwidth are king. The RX 9070 GRE uses 12 GB of GDDR6 with a 192-bit bus and 432.0 GB/s bandwidth, which is typical for a gaming card where latency and cost matter more than raw throughput. The MI60's memory clock is listed at 1000 MHz (2 Gbps effective), while the RX 9070 GRE runs at 2250 MHz (18 Gbps effective). The MI60's 4096-bit bus compensates for its slower memory clock, but the RX 9070 GRE's higher clock speed is what keeps it competitive in gaming scenarios.

Shader configuration also diverges. The MI60 has 4096 shading units, 256 TMUs, and 64 ROPs. The RX 9070 GRE has 3072 shading units, 192 TMUs, and 96 ROPs. Despite having fewer shaders and TMUs, the RX 9070 GRE achieves higher throughput because of its much higher clock speeds: its texture rate of 535.7 GTexel/s exceeds the MI60's 460.8 GTexel/s, and its pixel rate of 267.8 GPixel/s is more than double. The RX 9070 GRE also has 48 ray tracing cores, a feature absent from the MI60, which is critical for modern games that use hardware-accelerated ray tracing.

Power and connectivity also differ. The MI60 has a 300 W TDP and requires a 700 W power supply, using a 1x 6-pin + 1x 8-pin connector. The RX 9070 GRE has a 220 W TDP and recommends a 550 W PSU, using 2x 8-pin connectors. The MI60 uses PCIe 4.0 x16, while the RX 9070 GRE uses PCIe 5.0 x16, doubling the bandwidth available for data transfer. The MI60 has a single mini-DisplayPort 1.4a output, while the RX 9070 GRE has 1x HDMI 2.1b and 3x DisplayPort 2.1a outputs, making it far more flexible for multi-monitor setups. The MI60's dimensions are 267 mm (10.5 inches) in length and 111 mm (4.4 inches) in height, while the RX 9070 GRE's dimensions are not recorded in the database.

The Verdict

The data points to a clear split. The Radeon Instinct MI60 is for compute workloads that need large memory capacity and extreme bandwidth. Its 32 GB of HBM2 and 1.02 TB/s bandwidth are unmatched by the RX 9070 GRE, and its FP16 throughput of 29.49 TFLOPS is respectable for its era. It sits at the 93rd percentile among all GPUs, ahead of the NVIDIA RTX A4500 (which scores 91671, about 0.9% lower) and the RTX A4500 Mobile (91134, about 1.5% lower). It trails the AMD Radeon Pro VII (97131, about 4.8% higher) and the AMD Radeon RX 7900M (97487, about 5.2% higher), but those are closer in compute capability. The MI60 is end-of-life, so it is only relevant for existing systems or used purchases.

The Radeon RX 9070 GRE is for gaming and real-time graphics. Its 34.28 TFLOPS FP32, 267.8 GPixel/s pixel rate, and 48 ray tracing cores make it a far more capable gaming GPU. It supports modern APIs like DirectX 12 Ultimate and Vulkan 1.4, and its 12 GB of GDDR6 is adequate for most current games. Its average benchmark score of 57367 is skewed by the 3DMark Steel Nomad DX12 test (5424), but in the Geekbench OpenCL test it beats the MI60 by 15.4%. Its nearest rivals include the Intel Arc A580 (57756, about 0.7% higher), the AMD Radeon RX 5600 OEM (58085, about 1.2% higher), the Intel Arc A570M (58239, about 1.5% higher), and the AMD Radeon RX 6950 XT (58392, about 1.8% higher). These are close scores, suggesting the RX 9070 GRE is competitive with mid-range to upper-mid-range GPUs.

For a PC builder, the choice is simple: if the task is rendering, simulation, or AI inference with large models, the MI60 wins on memory capacity and bandwidth. If the task is gaming, streaming, or any workload that benefits from high FP32 throughput and modern API features, the RX 9070 GRE is the stronger pick. There is no overlap in their intended use cases, and the benchmark data confirms that.

FAQ

Q: Which card has higher raw compute performance?

A: The RX 9070 GRE has 34.28 TFLOPS FP32, more than double the MI60's 14.75 TFLOPS. In FP16, the RX 9070 GRE also achieves 34.28 TFLOPS (1:1), while the MI60 reaches 29.49 TFLOPS (2:1) but with reduced precision.

Q: Which card has more memory?

A: The MI60 has 32 GB of HBM2, while the RX 9070 GRE has 12 GB of GDDR6. The MI60 also has a 4096-bit bus and 1.02 TB/s bandwidth, versus the RX 9070 GRE's 192-bit bus and 432.0 GB/s bandwidth.

Q: Does the RX 9070 GRE support ray tracing?

A: Yes, it has 48 dedicated ray tracing cores. The MI60 has no ray tracing cores, so it cannot accelerate ray-traced workloads in hardware.

Q: What API level does each card support?

A: The RX 9070 GRE supports DirectX 12 Ultimate (12_2) and Vulkan 1.4. The MI60 supports DirectX 12 (12_1) and Vulkan 1.3. Both support OpenGL 4.6.

Q: Which card is more power-efficient?

A: The RX 9070 GRE has a 220 W TDP and recommends a 550 W power supply, while the MI60 has a 300 W TDP and recommends a 700 W power supply. The RX 9070 GRE also uses a smaller die (357 mm² vs 331 mm²) but with far higher transistor density (151.0M / mm² vs 40.0M / mm²).

Q: What is the release timeline for these cards?

A: The MI60 was released on 2018-11-17 and is now end-of-life. The RX 9070 GRE was released on 2025-05-07 and is still active. The RX 9070 GRE has a launch MSRP of 549 USD.

Head-to-Head Benchmarks

The only direct benchmark shared between the two cards is Geekbench OpenCL. The RX 9070 GRE scored 109309, while the MI60 scored 92488, giving the RX 9070 GRE a 15.4% lead. That result aligns with the architectural differences: the RX 9070 GRE's higher FP32 throughput (34.28 TFLOPS vs 14.75 TFLOPS) and much higher boost clock (2790 MHz vs 1800 MHz) give it a clear advantage in compute tasks that are not memory-bound. However, the MI60's 1.02 TB/s bandwidth and 32 GB capacity mean that workloads which fit into its memory hierarchy could see different results. The Geekbench test is likely not saturating memory bandwidth, so the MI60's main strength is not fully exercised.

The MI60's average benchmark score of 92466 is higher than the RX 9070 GRE's average of 57367, but that average includes a 3DMark Steel Nomad DX12 result for the RX 9070 GRE (5424) that has no counterpart for the MI60. If we isolate the Geekbench OpenCL results, the RX 9070 GRE is clearly ahead. The MI60's nearest rivals in the database are the NVIDIA RTX A4500 (91671, 0.9% lower), the RTX A4500 Mobile (91134, 1.5% lower), the AMD Radeon Pro VII (97131, 4.8% higher), and the AMD Radeon RX 7900M (97487, 5.2% higher). These are all data center or high-end mobile GPUs, reinforcing the MI60's compute positioning. The RX 9070 GRE's nearest rivals are the Intel Arc A580 (57756, 0.7% higher), the AMD Radeon RX 5600 OEM (58085, 1.2% higher), the Intel Arc A570M (58239, 1.5% higher), and the AMD Radeon RX 6950 XT (58392, 1.8% higher). These are gaming GPUs, and the RX 9070 GRE sits near them in average score, though its Geekbench OpenCL result is much stronger than its 3DMark result.

The pixel rate difference is stark: the RX 9070 GRE's 267.8 GPixel/s is 132% higher than the MI60's 115.2 GPixel/s. Texture rate is closer, with the RX 9070 GRE at 535.7 GTexel/s versus the MI60's 460.8 GTexel/s, a 16% gap. The MI60's ROP count of 64 is lower than the RX 9070 GRE's 96, yet the MI60's texture rate is still respectable because of its 256 TMUs. The RX 9070 GRE's 192 TMUs are fewer, but they run at much higher clocks.

Specification Differences

The two cards differ in nearly every specification that matters. The MI60 uses the Vega 20 chip (GCN 5.1) on a 7 nm process, with 13,230 million transistors and a 331 mm² die. The RX 9070 GRE uses the Navi 48 chip (RDNA 4.0) on a 4 nm process, with 53,900 million transistors and a 357 mm² die. Transistor density is 40.0M / mm² for the MI60 and 151.0M / mm² for the RX 9070 GRE.

Clocks: the MI60 has a base clock of 1200 MHz and a boost clock of 1800 MHz, with no game clock listed. The RX 9070 GRE has a base clock of 1420 MHz, a boost clock of 2790 MHz, and a game clock of 2220 MHz. Memory clocks are 1000 MHz (2 Gbps effective) for the MI60 and 2250 MHz (18 Gbps effective) for the RX 9070 GRE.

Memory: the MI60 has 32 GB of HBM2 on a 4096-bit bus, delivering 1.02 TB/s. The RX 9070 GRE has 12 GB of GDDR6 on a 192-bit bus, delivering 432.0 GB/s. The MI60's memory capacity is 2.67x larger, and its bandwidth is 2.36x higher.

Compute units: the MI60 has 4096 shading units, 256 TMUs, and 64 ROPs. The RX 9070 GRE has 3072 shading units, 192 TMUs, and 96 ROPs. The RX 9070 GRE also has 48 ray tracing cores, while the MI60 has none. FP32 output is 14.75 TFLOPS for the MI60 and 34.28 TFLOPS for the RX 9070 GRE. FP16 output is 29.49 TFLOPS (2:1) for the MI60 and 34.28 TFLOPS (1:1) for the RX 9070 GRE. Pixel rate is 115.2 GPixel/s for the MI60 and 267.8 GPixel/s for the RX 9070 GRE. Texture rate is 460.8 GTexel/s for the MI60 and 535.7 GTexel/s for the RX 9070 GRE.

Power and physical: the MI60 has a 300 W TDP, recommends a 700 W PSU, and uses a 1x 6-pin + 1x 8-pin connector. The RX 9070 GRE has a 220 W TDP, recommends a 550 W PSU, and uses 2x 8-pin connectors. Both are dual-slot cards. The MI60 is 267 mm long and 111 mm tall; the RX 9070 GRE's dimensions are not recorded.

Interface and outputs: the MI60 uses PCIe 4.0 x16 and has one mini-DisplayPort 1.4a. The RX 9070 GRE uses PCIe 5.0 x16 and has 1x HDMI 2.1b plus 3x DisplayPort 2.1a. API support: the MI60 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The RX 9070 GRE supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Production status: the MI60 is end-of-life, released on 2018-11-17. The RX 9070 GRE is active, released on 2025-05-07, with a launch MSRP of 549 USD.